瀏覽軌跡與語意解構中國山水

Scan-paths and semantic deconstruction of Chinese Landscape paintings

在實證美學研究中，觀看藝術作品主要探討的都是以凝視點及持續凝視時間為判斷畫中感興趣區域的主要依據之一。山水畫的圖像特徵會使眼睛運動產生大量的探索行為，凝視點的分布是呈現分散且不集中的現象，因此不容易被解讀。本研究以移動式眼動儀(MET)結合放聲思考的研究方式來分析觀看中國山水畫，在解構觀看歷程的同時，也建構我們對欣賞中國山水畫的行為認識；本研究選擇三幅不同形式的中國山水畫為受測樣本，並且在一個接近真實觀看藝術作品的展示空間中進行資料收集，將瀏覽軌跡和放聲思考的口述內容整合成具有語意意義的瀏覽軌跡，接著使用R語言找出瀏覽軌跡資料中的規則並且進行分類，最後將分類規則視覺化呈現。研究結果發現：三幅作品有各自獨特觀看的方式，分別為：集中注意型、圖文活躍型、繁複細節型等三種類型；本研究也發現，雖然畫面的繁複程度會影響瀏覽軌跡的規則數量，但圖文組合的作品中包含多種訊息，而使注意力活動範圍廣泛，瀏覽軌跡也呈現活躍的狀態，並且在語意表達上展現更豐富的層次。

In empirical aesthetic research, viewing works of art mainly discusses the point of gaze and the duration of gaze as a primary basis for judging the area of interest in the painting. The image characteristics of landscape painting will cause much exploratory eye movement behavior. The distribution of gaze points is scattered and not concentrated, so it is not easy to interpret. This research uses the mobile eye tracker (MET) combined with the research method of thinking aloud to analyze the viewing of Chinese landscape paintings. While deconstructing the viewing process, it also constructs our behavioral understanding of appreciating Chinese landscape paintings; This study selects three Chinese landscape paintings of different forms as the test samples, and collects data in a display space that is close to the actual viewing of artworks, integrate the scan-paths and the oral content of thinking aloud into a semantically meaningful scan-paths, then use the R programming language to find the rules in the scan-paths data and classify them, and finally visualize the classification rules. The results of the study found that: the three works have their unique ways of viewing, which are: focused attention, graphically active, and complicated details; this study also found that although the complexity of the screen affects the number of rules in the scan-paths, the combination of pictures and texts contains a variety of messages will not cause excessive distraction, nor will it generate too many scan-paths, but it will show a richer level of semantic expression.